Disaster Mitigation of Debris Flows, Slope Failures and Landslides
553
Impact of Forest Harvesting on Slope Stability in Managed Forest Catchment,
Nara Prefecture, Japan
Fumitoshi Imaizumi1) and Roy C. Sidle2)
1) Graduate School of Life and Environmental Science, University of Tsukuba, 1621-2, Ikawa University Forest,
Ikawa, Aoi, Shizuoka, 428-0504, Japan ([email protected])
2) Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto, Japan
([email protected])
Abstract
Forest harvesting that induces the root decay of logged trees and the root recovery of replanted trees
affects stability of slopes over time; these factors influence the occurrence of shallow landslides. Landslides cause
loss of life and property; thus, the occurrence of landslides is considered to be one of the most detrimental
effects of forest harvesting. Although many studies have been carried out to evaluate the effect of forest
harvesting on the occurrence of landslides, the impact of harvesting differs in each study due to dissimilar site
conditions (i.e., geology, physiography, and climate). Thus, field data that can neglect the influence of these
complex conditions is necessary for testing simulation model predictions of the effects of forest harvesting on
landslide occurrences. Using aerial photograph investigations and field surveys, we examined the occurrence
of landslides with regard to forest harvesting in the Sanko catchment (Nara prefecture, central Japan), where
rotational management yields a mosaic of different forest ages over the years. In the Sanko catchment, geology
and slope gradient are rather uniformly distributed, allowing the evaluation of the forest harvesting effect by
ignoring other factors. The trends in the landslide frequency in the managed forests correspond to changes
in slope stability; this is explained by root strength decay and recovery. The impact of landslide occurrence
can be summarized to be strongest for 1–10 yr after forest harvesting and smaller impacts continue up to 25
yr after harvesting. In the Sanko catchment, the influence of forest harvesting on landslide occurrence is more
evident in steep slopes as compared to gentler slopes. Only smaller landslides (< 100 m3 ) whose slide surface
might be shallower than root depth are accelerated by forest harvesting, while the effect of forest harvesting
is not obvious for larger landslides (> 100 m3 ). These characteristics of landslide frequency relevant to forest
harvesting can be demonstrated by landslide models based on infinite slope stability.
Keywords: landslide, forest harvesting, slope stability, root strength
Introduction
Acceleration of landslides with destructive power instantaneously supplies enormous volumes of sediment into streams. These landslides are considered to be the most significant impact of the forest harvesting.
An increase in the frequency of landslides may cause disasters in mountainous regions. Thus, the effect of
forest harvesting, including clear cutting and forest regeneration, on the occurrence of landslides needs to be
investigated in order to develop better measures for mitigating disasters.
The quantification of the effects of forest harvesting on landslide rates has been conducted, particularly
in the Pacific Northwest (Jacob, 2000; Brardinoni et al, 2002; Guthrie, 2002). Although these studies revealed
that clear cutting accelerates the occurrence of landslides, the impact of logging differs in each study. For
example, the number and volume of landslides in clear cut terrains is more than 10-fold higher than that in
undisturbed forests in some catchments, while the effect of logging on the occurrence of landslides is unclear
in other catchments located in the same region (Jacob, 2000; Brardinoni et al, 2002; Guthrie, 2002). These
studies evaluated the effects of forest harvesting by comparing the landslide frequency (or volume) in logged
forests with that of undisturbed forests located in other areas. However, not only forest harvesting but also
other factors (i.e., geology, physiography, and climate) affect the occurrence of landslides (Sidle et al., 1985;
Brardinoni et al, 2002). Thus, the influence of forest harvesting on the occurrence of landslides in certain
terrains cannot be elucidated by comparing the landslide frequency with the different geology, physiography,
and climate.
The overall aim of this study is to clarify the effects of forest harvesting (i.e., clear cutting and
subsequent regeneration of artificial forests) on the occurrence of landslides in the Sanko catchment, central
Japan. In the Sanko catchment, the geology and slope gradient are rather uniformly distributed, thereby
c
pp. 553–560 °2006
by Universal Academy Press, Inc. / Tokyo, Japan
554
Fig. 1.
Topographic map of Sanko catchment.
enabling the evaluation of the effect of forest harvesting without considering the other factors. The specific
objectives include: (i) assessing the occurrence of landslides in the managed forest, (ii) examining the effects
of forest harvesting on the occurrence of landslides, and (iii) investigating the influence of forest harvesting on
hillslope stability based on physical analysis.
Study area
Site description
The Sanko catchment is drained by Kanno River, a headwater tributary of the Kumano River; this
catchment is located in the southwestern Nara prefecture, central Japan (Fig. 1). This catchment has an area
of 8.50 km2 with tributaries flowing in the north-south direction. The lowest point of the catchment is located
in the southeast end of the catchment (750 m a.s.l.) and the highest point is the peak of Mount Gomadan (1372
m a.s.l.) in the southwest region of the catchment. The area is underlain by the Cretaceous Shimanto belt
comprised of sandstone and clay, and the area is homogeneous throughout the catchment. Hillslope gradients
are also relatively homogeneous throughout the catchment (typical gradients 30◦ –50◦ , Fig. 1); thus, the effect
of differences in the hillslope gradient on the distribution of landslides may not be significant. The channel
gradient is 2◦ –5◦ for the main stream (Kanno River) and 5◦ –35◦ for tributaries. The soils are shallow, typically
ranging 0.5–1.0 m, because of the steep hillslopes.
An annual rainfall of 2500 mm occurs at the Wakayama Experimental forest of Kyoto University,
about 3 km west of the Sanko catchment. Heavy rainfall events (i.e., total storm rainfall > 100 mm) occur
during the Baiu rainy season (June and July) and in the autumn typhoon season (from late August to early
October). Winter snowfall occurs at higher elevations within the catchment, but precipitation from December
to February is only about 10% of the total annual precipitation. Thus, snowmelt is typically not a significant
landslide-triggering mechanism in this area.
Forest management
About 95% of the Sanko catchment has been converted into an artificial forest (largely sugi (Japanese
cedar) with minor amounts of hinoki (Japanese cypress) and the remainder comprises secondary broadleaf
555
Fig. 2.
The year of replanting in Sanko catchment.
forests, forest roads, and log landings. Broadleaf forests and log landings are mainly located on mountain
ridges. In this study, we did not compare the occurrence of landslides in broadleaf forests and log landings
with those in artificial forests because of their different geographic conditions. Clear cutting has been the only
harvesting method used in the catchment. Typically, replanting is carried out one or two years after logging.
In the Sanko catchment, both sugi and hinoki forests are cultivated in rotation in intervals of about 80 yrs;
thus, regenerating forests with different ages (representing different periods of clear cutting) exist during years
of abundant as well as scanty rainfall (Fig. 2). By averaging landslides rates within the various age classes of
forest stands over long periods we can evaluate the effects of forest harvesting with the influence of rainfall
minimized. Because only skyline logging has been conducted in the Sanko catchment, the influence of logging
methods on landslide occurrence (e.g., Roberts et al., 2004) cannot be expected. Old-growth artificial forests
(both sugi and hinoki), which were replanted from 1912–1916, are investigated as a control area spanning across
0.78 km2 (about 9% of the entire catchment). The influences of logging and replanting appear to be minimal
in the control area because the forest ages from 1964–2002 are 38–90 yr.
Methodology
Monochrome aerial photographs for nine periods (1964, 1965, 1967, 1971, 1984, 1989, 1994, 1998,
2003) and color aerial photographs for 1976 were used to assess the location and area of landslides in the Sanko
catchment. These landslides were confirmed by stereo photograph pairs and were mapped on 1:5000 forest
management maps. Most of the aerial photographs were taken in March (before the Baiu season); thus, almost
all the mass movements confirmed by aerial-photo stereographs probably occurred prior to December of the
previous year. New occurrences of landslides for each period (1964, 1965–1966, 1967–1970, 1971–75, 1976–
1983, 1984–1993, 1994–1997, 1998–2002) were confirmed by comparing earlier and later aerial photographs.
Expanded landslides were not considered to be new landslides in this study. Mapped landslides were scanned
and their location and areas were analyzed using Arc GIS software.
Volumes of 11 landslide scars, including their initiation and transportation zones, were measured in the
field to develop a preliminary volume-area relationship for landslides within the catchment. The landslide size
ranged from 50–4,000 m2 , thereby including most landslide sizes in the catchment, as confirmed by the aerial
photograph investigations. This relationship was used to estimate the landslide volume from the landslide area
obtained from the aerial photograph interpretations. To estimate volume of sediment supply from expanded
landslides, volume of the landslide in the prior period was subtracted from volume of the later landslides.
Result
A total of 873 landslides were confirmed from the aerial photograph investigations in the period from
1964–2002. Since 256 landslides were confirmed from the aerial photographs obtained in 1964 (the first period),
617 new landslides occurred from 1964–2002. In the Sanko catchment, forest roads exist mainly along the
mountain ridgelines and in the valley bottom along the stream. These roads sometimes have the greatest
556
Fig. 3. Changes in sediment supply rate from new/expanded landslide with increase in duration
between forest harvesting and occurrence of landslides.
Fig. 4. Changes in root strength estimated from Sidle (1992). Maximum root strength = 10,000 Pa
was assumed. Root regrowth curve estimated from equation [1] and root decay curve estimated
from equation [2] are also illustrated in the figure.
special impact on landslide initiation (Wemple et al., 2001; Brardinoni et al., 2002; Grthrie, 2002), and 124
landslides were initiated from roads in the Sanko catchment in the period from 1964 to 2002. The landslides
initiating from roads were excluded from assessments in this study in order to clarify the impact of clear
cutting and subsequent forest regeneration on landslide occurrence. Aerial photograph investigations cannot
confirm smaller landslides, and the threshold scale of non-visible landslides depends on forest cover conditions
(Brardinoni et al, 2003; Brardinoni and Church, 2003). Sugi and hinoki (both conifers) are typically replanted
in an evenly distributed pattern in the study catchment, thus facilitating the recognition of smaller landslides
in aerial photographs. Intensive field surveys conducted in some sub-catchments revealed that landslides with
sizes exceeding approximately 50 m2 are likely to be detected in aerial photographs; thus, landslide sizes
documented in our study range approximately from 50–5000 m2 . Based on both aerial photographs and field
investigations, landslides typically occur at the bedrock surface or shallower depths at an average depth of only
0.6 m. However, only a few landslides are associated with bedrock failure.
In order to help quantify the effect of forest harvesting on the occurrence of landslides, the duration
between forest harvesting at headwalls of landslides and their occurrence were investigated using GIS software.
Sediment supply rate from the new/expanded landslides — calculated from the annual volume of new/expanded
landslides divided by the area of each duration categories — is the largest in forests 6–10 yr after harvesting
(Fig. 3).The sediment supply from landslides in the forests 1–10 yr after harvesting is about nine times higher
compared to the control sites. The sediment supply in the forests more than 26 yr after harvesting is similar
to that of the control sites, indicating that slope stability almost completely recovered within 25 yr after
harvesting.
557
Fig. 5.
Schematic diagram of slope stability.
Discussion
Influence of root strength on slope stability
Sidle (1991) quantified the general shape of a conceptual root strength regrowth curve using the
following equation:
R = (a + be−kt )−1 + d
(1)
where R (dimensionless) is the actual root cohesion divided by the maximum root cohesion (dimensionless), t
is the time elapsed since the harvesting years (yr), and a, b, d, and k are the empirical constants. The empirical
coefficients based on the uprooting tests for sugi trees are a = 0.952, b = 19.05, d = −0.050, and k = 0.250
(Sidle, 1991). Sidle (1992) predicted the root strength decline (D) using the following equation:
D = e−ltn
(2)
where D is the actual root cohesion divided by the maximum root cohesion (dimensionless) and l and n are
the empirical constants. Unfortunately, the decay coefficients for sugi and hinoki were unavailable. However,
the decay coefficients for coastal Douglas firs (l = 0.506 and n = 0.730; Sidle, 1992) with root strength decline
trends similar to sugi (Tsukamoto, 1987) can be substituted for those of sugi. The changes in R and D in the
Sanko catchment are estimated using equations [1] and [2] and the empirical coefficients of Sidle (1991, 1992).
The changes in the total actual root cohesion after harvesting are estimated from the sum of R and D multiplied
by the estimated maximum root strength (= 10,000 Pa). The changes in hillslope instability assumed from the
changes in the frequency of landslides (Fig. 3) are antithetical to the changes in the estimated net root strength
(Fig. 4). Thus, the temporal changes in frequency of landslides in the Sanko catchment can be explained by
the root strength models proposed by Sidle (1991, 1992).
Inference of slope gradient on slope stability
The factor of safety (FS) for slope stability is the ratio of shear strength to shear stress and is given
by
Rh
Rh
c + c0 + [ hw {(1 − n)γs + nSγw }dz + z1w (1 − n)(γs − γw )dz]g cos2 θ tan φ
FS =
R hw
{(1 − n)γs + nSγw }dzg cos θ sin θ
z1
(3)
where c is the effective soil cohesion (Pa), c0 is the cohesion attributed to the root system (Pa), n is the porosity
of soil (dimensionless), S is the degree of saturation (dimensionless), γs is the mass density of sediment (or soil)
particles, γw is the mass density of water, θ is the slope gradient, φ is the effective internal angle of friction
(degree), z1 is the height of slide surface from the reference plane; hw is the height of ground water surface
from the reference plane, h is the height of slope surface (Fig. 5). Under the assumption that parameters in
equation [3] are constant along depth direction, the following values estimated for each parameter were used for
investigating the effect of slope gradient and root strength on slope stability in the Sanko catchment: c = 3000
(Pa), n = 0.4, S = 1, γs = 2600 (kg /m3 ), γw = 1000 (kg /m3 ), g = 9.8 (m/s2 ), φ = 38 (degree), h − z = 0.6
(m), and hw − z = 0.6 (m). The average root strength from 0–10 yr (when slopes are the most unstable in the
Sanko catchment in Fig. 3), 11 to 25 yr (when the slope stability is recovering), and greater than 25 yr after
harvesting (when the slope stability has almost recovered) in Fig. 4 were used to calculate the slope stability
using equation [3] (Fig. 6). Parameters in the site exhibit variance around the mean values and are distributed
throughout the catchment. Although the calculated FS always exceeds 1 (Fig. 6), landslides possibly occur on
558
Fig. 6.
Changes in the factor of safety with increasing in slope gradient.
Fig. 7.
Contribution of root strength to slope stability.
the slopes with negative parameters with regard to slope stability as compared to the mean values. For any
duration considered after harvesting in Fig. 6, the FS decreases and approaches 1 with an increase in the slope
gradient. This indicates that steeper slopes are more unstable compared to gentler slopes if parameters with
the exception of slope gradient are similar. Differences in FS for each duration are greater in gentler slopes
(Fig. 6); however, the contribution of root strength to the total slope stability is not clear. The contribution
of root strength to the total slope stability of forests more than 25 yr after harvesting was calculated using
the root strength of the matured forest (≈ 10,000 Pa) divided by the sum of the numerator in the equation
[3]. Root strength in the numerator of the equation [3] was also assumed to be 10,000 Pa for the calculation.
The contribution of the root strength to the total slope stability of steeper slopes is greater as compared to
the gentler slopes (Fig. 7), indicating that the stability of the steeper slope is more sensitive to the changes in
the root strength caused by root decay and regrowth. In the Sanko catchment, increases in the frequency of
occurrence of new landslides — calculated from the number of new landslides per year divided by the area of
duration category — with slope gradient are not clear for forests 26–40 yr after harvesting (Fig. 8). However,
the frequency of occurrence of new landslides for forests 0–10 yr after harvesting increases with the slope
gradient, indicating that the influence of harvesting on landslide frequency is more evident for steeper slopes.
This larger effect of forest harvesting on slope stability of steeper terrains agrees with the larger contribution
of root strength to the total slope stability elucidated by infinite slope model (Fig. 7).
Landslide sizes induced by forest harvesting
Frequency of smaller landslides (< 100 m3 ) in the forests, 1–10 yr after harvesting, is more than 8-fold
greater than that in forests 26–40 yr after harvesting, while the frequency of larger landslides in the forests
1–10 yr after harvesting is only 3.4-fold greater in forests 26–40 yr after harvesting (Table 1). The area-depth
relationship of landslides developed by field surveys predictably shows that the depth of smaller landslides is
559
Fig. 8. Changesin frequency of new landslides for each volume category with increase in duration
between forest harvesting and occurrence of landslides.
Table 1.
Acceleration in frequency of occurrence of new landslides for each landslide volume category.
shallower than that of larger landslides. The average depth of smaller landslides (< 100 m3 ) measured in the
study site was about 40 cm lesser than the depth of sugi roots (50–60 cm for lateral roots, and 1 m for vertical
roots; Tsukamoto, 1987). Thus, the effect of forest harvesting on slope stability is sensitive at depths shallower
than those of tree roots.
Conclusions
As demonstrated in this study, changes in frequency of landslides in the Sanko catchment can be
explained by the root strength models proposed by Sidle (1991, 1992). This indicates that the landslide
frequency trends are greatly affected by changes in root strength in managed forests. Magnitude and duration
of the influence of forest harvesting differ depending on both site conditions (i.e., geology, physiography, and
climate) and tree conditions (i.e., species and environmental conditions for tree growth; Sidle, 1991; Sidle,
1992). Analysis of infinite slope stability can demonstrate that the acceleration of landslide occurrence induced
by forest harvesting is more evident in steeper terrains; thus, landslide models may be useful for estimating the
effect of physiography on the occurrence of landslides. However, information on other factors that affect the
occurrence of landslides is scarce, particularly information related to tree conditions. In order to estimate the
magnitude and duration of forest harvesting on slope stability in other regions, information for other factors
should be investigated.
References
Brardinoni, F., Hassan, M. A., and Slaymaker, (2002) Complex mass wasting response of drainage basins to
forest management in coastal British Colombia, Geomorphology. 49, p.109–124
Gurhrie, R.H. (2002) The effects of logging on frequency and distribution of landslides in three watersheds on
Vancouver islands, Geomorphology. 43, p. 275–294
Roberts, B., Ward, B., and Rollerson, T. (2004) A comparison of landslide rates following helicopter and
conventional cable-based clearcutting operations in the Southwest Coast Mountains of British Columbia,
Geomorphology. 61, p. 337–346
Brardinoni, F., Hassan, M. A., and Slaymaker, (2002) Complex mass wasting response of drainage basins to
forest management in coastal British Colombia, Geomorphology. 49, p. 109–124
Brardinoni, F., Slaymaker, O., Hassan, M. A. (2003) Landslide inventory in a rugged forested watershed: a
comparison between air-photo and field survey data, Geomorphology. 54, p. 179–195
560
Brardinoni, F., and Church, M. (2004): Representing the landslide magnitude-frequency relation: Capilano
River Basin, British Columbia, Earth Surf. Process. Landforms. 29, p. 115–124.
Gurhrie, R.H. (2002) The effect of logging on frequency and distribution of landslides in three watersheds on
Vancouver islands, Geomorphology. 43, p. 275–294
Jakob, M., Bovis, M., and Oden, M. (2005): The significance of channel recharge rates for estimating debris-flow
magnitude and frequency, Earth Surf. Process. Landforms. 30, p. 755–766
Sidle, R. C., Pearce, A. J., and O’Loughlin, C. L. (1985) Hillslope stability and land use, American Geophysical
Union Water Resources Monograph 11, 140p.
Sidle, R.C. (1991). A conceptual model of changes in root cohesion in response to vegetation management. J.
Environ. Qual. 20, p. 43–52
Sidle. R.C. (1992). A theoretical model of the effects of timber harvesting on slope stability, Water Resour.
Res. 28. p. 1897–1910.
Tsukamoto, Y. (1987) Evaluation of the effect of tree roots on slope stability, Bulletin of the Experiment Forests,
Tokyo University of Agriculture and Technology. 23, p. 65–124 (in Japanese with English abstract).
Wemple, B. C., Swanson, and F., Jones, A. A. (2001) Forest roads and geomorphic processes interactions
cascade range, Oregon, Earth Surf. Process. Landforms. 26, p. 191–204